In bacteria, Dps is one of the critical proteins to build up a condensed nucleoid in response to the environmental stresses. In this study, we found that the expression of Dps and the nucleoid condensation was not simply correlated in Escherichia coli, and that Fis, which is an E. coli (gamma‐Proteobacteria)‐specific nucleoid protein, interfered with the Dps‐dependent nucleoid condensation. Atomic force microscopy and Northern blot analyses indicated that the inhibitory effect of Fis was due to the repression of the expression of Topoismerase I (Topo I) and DNA gyrase. In the Δfis strain, both topA and gyrA/B genes were found to be upregulated. Overexpression of Topo I and DNA gyrase enhanced the nulceoid condensation in the presence of Dps. DNA‐topology assays using the cell extract showed that the extracts from the Δfis and Topo I‐/DNA gyrase‐overexpressing strains, but not the wild‐type extract, shifted the population toward relaxed forms. These results indicate that the topology of DNA is dynamically transmutable and that the topology control is important for Dps‐induced nucleoid condensation.
It is intriguing that very long genomic DNA molecules ranging from ∼cm to ∼m (corresponding to ∼106 to ∼109 base pairs) are stored in small containers with ∼μm in diameter. Two strategies have been taken in the evolution of life to accommodate the genomes in cells: in bacteria, the genomic DNA is packed in a cell as a form of ‘nucleoid’ (Robinow and Kellenberger, 1994; Poplawski and Bernander, 1997; Azam et al, 2000), whereas, in eukaryotic cells, the genomic DNA exists in a form of chromatin and is packed in a nucleus (Kornberg, 1974; Thoma et al, 1979; Widom and Klug, 1985). In either case, to organize the DNA into higher‐order structures, a set of distinct structural DNA‐binding proteins, such as histones in eukaryotic cells and Hu in bacteria, constitutively play major roles by utilizing the physical/chemical properties of DNA–protein interactions. A number of additional proteins, such as SMC (structural maintenance of chromosome) proteins and topoisomerases, also play crucial roles in the construction, maintenance and re‐construction of well‐organized higher‐order structures of genomes (Hayat and Mancarella, 1995; Swedlow and Hirano, 2003).
The formation of the higher order structures itself has a biological significance in the protection of the genomic DNA against the environmental stresses. In eukaryotes, DNA damage caused by UV light or oxidative stress accumulate less in nucleosomes than in naked DNA (Ljungman, 1991; Ljungman and Hanawalt, 1992; Yoshikawa et al, 2006). Furthermore, the formation of higher order chromatin dramatically decreases the DNA damage (Ljungman, 1991; Ljungman and Hanawalt, 1992).
In bacteria, one of the proteins to transform the nucleoid into condensed state is Dps (DNA‐binding protein from starved cells) (Almiron et al, 1992). Dps is a stress‐induced protein with a molecular weight of 19 kDa and is known to be a member of the Fe‐binding protein family that forms dodecameric complex in cells (Grant et al, 1998). Dps protects genomic DNA against oxidative stress (Martinez and Kolter, 1997), nuclease cleavage, UV light, thermal shock (Nair and Finkel, 2004) and acid (Choi et al, 2000), possibly by its DNA‐binding ability to block the stress elements that attack DNA. Its Fe‐chelating activity is also the important feature in the oxidative stress resistance, because iron (Fe2+) supplies an electron to produce the hydroxyl radicals via Fenton reaction (Zhao et al, 2002) and these radicals damage various critical macromolecules including the genomic DNA (Nunoshiba et al, 1999). Dps can reduce the intracellular level of Fe2+ and thus restricts the production of hydroxyl radicals (Zhao et al, 2002).
Interaction between DNA and Dps results in a DNA–Dps co‐crystal both in vitro and in vivo. In vitro, the mixture of naked DNA and Dps rapidly induces the crystalline state (Wolf et al, 1999; Frenkiel‐Krispin et al, 2001). In Escherichia coli, Dps becomes the most abundant nucleoid component in the stationary phase (Azam and Ishihama, 1999) and causes a condensation of nucleoid (Wolf et al, 1999; Frenkiel‐Krispin et al, 2001; Kim et al, 2004). Electron microscopy observations have revealed that the nucleoid in the stationary phase forms the biocrystal that is composed of the toroidally assembled Dps and DNA (Wolf et al, 1999; Frenkiel‐Krispin et al, 2001, 2004). Even after the lysis of the cell, the nucleoid in the stationary phase is in a tightly condensed state (Kim et al, 2004).
The effect of Dps on the nucleoid state can vary depending on the growth conditions or on the bacterial species. The overexpression of Dps in the log phase E. coli induces neither nucleoid condensation nor DNA–Dps co‐crystallization (Frenkiel‐Krispin et al, 2001), whereas Dps overexpression in the stationary phase caused nucleoid condensation (Kim et al, 2004). In contrast, the nucleoid in Staphylococcus aureus transforms its structure into a condensed state in both log and stationary phases by the induction of mrgA, a staphylococcal ortholog of dps (Morikawa et al, 2006). In addition, in S. aureus, the oxidative stress promotes the expression of MrgA and nucleoid condensation (Morikawa et al, 2006), whereas, in E. coli, as shown in this article, an induction of Dps by oxidative stress in the log phase did not result in the condensation of the nucleoid. Based on these differences between E. coli and S. aureus, we can postulate the presence of additional factors regulating the nucleoid condensation in E. coli.
In this study, we show that Fis is the factor inhibiting the nucleoid condensation in the log phase. Fis is abundant in the log phase and is one of the major DNA‐binding nucleoid proteins (Azam and Ishihama, 1999). It also acts as a transcription factor (Nilsson et al, 1990; Ross et al, 1990; Xu and Johnson, 1995; Hengen et al, 1997) to regulate a variety of genes. Our analyses show that Fis is one of the specific protein in gamma‐Proteobacteria including E. coli and is not present in S. aureus, and that topoisomerase I (TopoI) and DNA gyrase, which are under the control of Fis in E. coli, facilitate the nucleoid condensation. Since Fis, Topo I and DNA gyrase can change the DNA superhelicity (Schneider et al, 1997; Schneider et al, 1999; Weinstein‐Fischer et al, 2000), we propose that the control of DNA topology by these proteins is critical for the Dps‐induced nucleoid condensation.
Dps expression in E. coli
Two different transcription factors are involved in the dps‐gene regulation in E. coli (Figure 1A). First, IHF controls the accumulation of Dps in the stationary phase (Altuvia et al, 1994; Lomovskaya et al, 1994). Second, OxyR, an LysR‐family transcription regulator, upregulates the dps gene under oxidative stress (such as in the presence of hydrogen peroxide (H2O2)) (Altuvia et al, 1994; Lomovskaya et al, 1994). Our Western blot analysis of wild‐type (wt) E. coli showed that the levels of Dps expression in the stationary phase and under the oxidative stress were about 10 times and three times higher than that in the log phase, respectively (Figure 1B). On the other hand, in the Δdps strain lacking the dps gene, Dps was never detected under all the conditions examined (Figure 1B). In the ΔoxyR strain that lacks the oxyR gene, the expression of Dps stayed unchanged even after exposure to the oxidative stress, but was highly induced in the stationary phase (Figure 1B). A deletion of ihf (ΔhimD) diminished the induction of Dps during the transition into the stationary phase, but it still allowed the Dps induction under oxidative stress (Figure 1B). Thus, we conclude that the mechanisms of transcriptional regulation of dps gene are intact and physiologically functional in our experimental system.
Correlation between the Dps expression and the nucleoid condensation
We have developed ‘on‐substrate lysis’ procedure (see Materials and methods) for removing the cellular membrane and subsequent direct observation of the bacterial nucleoid and eukaryotic chromatin in cells by atomic force microscopy (AFM) (Yoshimura et al, 2003; Kim et al, 2004). This procedure can be applicable to evaluate the efficiency of nucleoid condensation in E. coli, because not‐condensed nucleoid extends fibrous structures around the lysed cell, whereas the condensed nucleoid cannot release fibers upon lysis (Kim et al, 2004; Morikawa et al, 2006). In this study, to elucidate the regulatory factors involved in the Dps‐induced nucleoid condensation in E. coli, we first assessed the degree of nucleoid condensation under various conditions that induced the Dps expression.
After ‘on‐substrate lysis’, the state of the nucleoid was observed by 4′,6‐diamino‐2‐phenylindole (DAPI) staining. The number of DAPI‐stained cells whose nucleoids remained compacted or spread out of the cell (termed ‘lysed’) was counted, and used as an indication of the degree of the nucleoid condensation. In the log phase, 95 and 93% of the wt and Δdps cells, respectively, were lysed and appeared to have a non‐condensed nucleoid (Figure 2A, B and H). Close observations by AFM demonstrated the existence of nucleoid fibers around the lysed cells (Figure 2C and I). In the stationary phase, only 12% of the wt cells were lysed (i.e. condensed) (Figure 2 A, F and G). Although 44% of the Δdps cells were not apparently lysed (Figure 2A), dispersed nucleoid fibers were still observed in the Δdps strain (Figure 2L and M). Namely, the nucleoid becomes condensed during the transition into the stationary phase in a Dps‐dependent manner.
In contrast, the oxidative stress given in the log phase induced the expression of Dps but did not condense the nucleoid. Eighty‐five percent of the wt cells were efficiently lysed even after the induction of Dps by 2 mM H2O2, and the nucleoid fibers were observed around the lysed cells (Figure 2A, D and E). Negative control strains, Δdps and ΔoxyR strains, exhibited 81 and 72% efficiency of lysis after the treatment with H2O2, respectively (Figure 2A, J, K, P and Q). Therefore, we concluded that the induction of Dps by oxidative stress does not induce the nucleoid condensation in E. coli. This implies, in turn, the existence of a log phase specific factor(s) that inhibits the nulceoid condensation under oxidative stress.
Role of Fis on the Dps‐induced nucleoid condensation
In E. coli, Fis is the most abundant nucleoid component in the log phase (∼60 000 molecules/cell), whereas its expression becomes undetectable at the stationary phase (Azam and Ishihama, 1999). When Dps is overexpressed in the wt (fis+) strain in the log phase, the nucleoid was not condensed (Kim et al, 2004). S. aureus, in which the expression of MrgA (Dps ortholog) is directly coupled with the nucleoid condensation (Morikawa et al, 2006), does not possess fis nor its homologous genes (Takeyasu et al, 2004). Therefore, we suspect Fis to be an inhibitory factor against the nucleoid condensation. To test this hypothesis, Dps was transiently overexpressed in the log phase of a Δfis strain of E. coli using a Dps‐expression plasmid, and possible changes of the nucleoid structure were examined by DAPI staining and AFM observation. As expected, the induction of Dps by isopropyl‐β‐d‐thiogalactopyranoside (IPTG) condensed the nucleoid even in the log phase, and the efficiency of cell lysis was decreased to 16% (Figure 3A, B and F) from the original value of 80% (Figure 3A and D). Close observations by AFM revealed that the Δfis strain exhibited the fibrous and condensed nucleoid in the absence and presence of Dps overexpression, respectively (Figure 3E and G).
We further investigated whether or not the nucleoid condensation could be induced in the Δfis strain by oxidative stress. Western blot analyses against Dps showed that a treatment of the Δfis strain with 2 mM H2O2 induced as much amount of Dps as in the wt treated with 2 mM H2O2 (Figure 3C). On‐substrate‐lysis treatment of the cells and the subsequent DAPI staining demonstrated that only 18% cells were lysed (Figure 3A and H). A close observation by AFM found that the nucleoid in the Δfis cells was condensed under H2O2 treatment (Figure 3I). In the stationary phase of the Δfis cells, Dps was normally expressed (Figure 3C), and the nucleoid was condensed (Figure 3A, J and K). Thus, we concluded that Fis is the inhibitory factor for the Dps‐dependent nucleoid condensation in the log phase, but it does not participate in the regulation of the dps‐gene expression.
Effects of Topo I and DNA gyrase on the Dps‐induced nucleoid condensation
Fis is known to work as a transcription regulator for the topA, gyrA and gyrB genes, each coding for Topoisomerase I (Topo I), DNA gyrase subunit A (GyrA) and subunit B (GyrB), respectively (Schneider et al, 1999; Weinstein‐Fischer et al, 2000). Indeed, our Northern blot analysis indicated that the expression of these genes were upregulated in the Δfis strain (Figure 4). Topo I is known to relax supercoiled DNA, whereas DNA gyrase causes negative supercoiling of DNA. Therefore, the repression of nucleoid condensation by Fis might be due to the changes in intracellular levels of Topo I, GyrA and GyrB, ultimately leading to the changes in DNA topology. This interpretation is in good agreement with the fact that the topology control of DNA is critical for achieving the higher‐order structures of DNA (Yoshimura et al, 2000; Hizume et al, 2004; Hizume et al, 2005). We examined this possibility by monitoring the in vivo status of DNA topology in the wt and Δfis strain. Agarose gel electrophoresis containing chloroquine showed that, in the log phase, the population distribution of supercoiled plasmids was larger in the wt than in Δfis strain, being consistent with our interpretation (Supplementary Figure 1).
We constructed a wt (fis+) strain that can transiently express His‐tagged Topo I, as described in Materials and methods. The induction of His‐Topo I expression by IPTG was evidenced by Western blot analysis (Figure 5A). In these His‐Topo I expressing cells, the nucleoid was easily dispersed as in the wt cells (Figure 5C). In contrast, when Dps was induced by the treatment with 2 mM H2O2 after Topo I induction, the nucleoid became highly condensed (Figure 5D–F). In the cells expressing His‐tagged DNA gyrase (GyrA and GyrB) (Figure 5B), the nucleoid without H2O2 was easily dispersed (Figure 5G), whereas the nucleoids were also condensed when treated with 2 mM H2O2 (Figure 5H–J).
Cell extracts from Δfis, Topo I++ and DNA Gyrase++ cells control DNA topology
The expression of gyrA and gyrB is induced by the decrease in negative supercoil in cells (Menzel and Gellert, 1987; Peter et al, 2004), and the increase in negative supercoil facilitates topA expression (Menzel and Gellert, 1983; Mizushima et al, 1993; Ogata et al, 1994). Thus, DNA topology can be controlled by Topo I and GyrA/GyrB in cells. Northern blot analyses indicated that the overexpression of GyrA and GyrB upregulated the expression level of Topo I, but that the overexpression of Topo I did not change the amount of gyrA/gyrB mRNA in our experimental system (Figure 6A–C). In any case, Topo I and DNA gyrase coexist in these overexpressing cells.
To clarify the biological significance, we further examined how efficiently different cell extracts derived from the wt, Δfis and Topo I‐/DNA gyrase‐overexpressing strains (Topo I++ and Gyrase++) could change the topology of DNA in vitro. The agarose gel electrophoresis containing chloroquine showed that, whereas wt extract never changed the population of topoisomers of DNA, the extract from Δfis, Topo I++ and Gyrase++ changed the population toward relaxed form (Figure 7A). These results indicate that the topology of DNA is dynamically transmutable in the Δfis, Topo I++ and Gyr ase++ strains, but not in the wt strain. Also, it is interesting to note that the cell extract from Gyrase++ had a very similar effect to that from Topo I++. It seems that the effect of Topo I in the extract predominate over the effect of DNA gyrase, and the effects of DNA gyrases in the extract may be very subtle and involved a local topology control. Indeed, when the extract from Gyrase++ was added to the extract from Topo I++, the topoisomers were not shifted toward supercoiling but toward relaxed form (Figure 7B, also see Supplementary Figure 2).
It is well known that DNA gyrase requires ATP and Mg2+ for its action. Consequently, we examined the effect of ATP (1∼10 mM) and Mg2+ (1∼100 mM) in the DNA gyrase++ cell extracts on the DNA topology. Unexpectedly, the populations of topoisomers were shifted toward the relaxed forms in a similar extent to the effect in the absence of ATP and Mg2+ (data not shown). It seems that the effect of Topo I predominates over the effect of DNA gyrase.
We have previously reported that, in S. aureus, the induction of Dps ortholog, MrgA, by oxidative stress induced a nucleoid condensation. The Dps‐dependent nucleoid condensation was postulated as the genome protection system against oxidative stress. However, in this study, we demonstrated that the oxidative stress did not simply lead to the nucleoid condensation in E. coli due to the effects of Fis as the regulator of the topA and gyrA/B gene expression that is required for the control of DNA topology.
DNA topology maintenance by Fis, Topo I and DNA gyrase
Fis preferably binds to the 15 bp consensus sequence, but when it exists in excess amount, its DNA‐binding becomes sequence nonspecific. In vitro, Fis changes the overall shape of supercoiled DNA in a sequence‐independent manner (Schneider et al, 2001; Hardy and Cozzarelli, 2005), and prevents the topological changes caused by DNA gyrase and Topo I (Schneider et al, 1997). In addition, our results showed that Fis repressed the expressions of gyrA, gyrB and topA (Figure 4). These facts suggest that Fis plays a role to sustain the DNA superhelicity at the steady‐state level using two different pathways: one is acting as a physical barrier and buffering the change in the DNA superhelicity, and the other is repressing the expression of the factors that change the DNA superhelicity (Figure 8A). In fact, the cell extract from the wt (fis+) strain could not affect the population distribution of topoisomers, whereas the cell extract from the Δfis strain changed the population distribution of the topoisomers (Figure 7).
The DNA topology should be always balanced in a cell. In this sense, all of the following may be critical: (i) direct binding of Fis to DNA and forming a physical barrier (Schneider et al, 1997, 2001; Hardy and Cozzarelli, 2005); (ii) controlling gene expression including topA and others (Schneider et al, 1999; Weinstein‐Fischer et al, 2000); (iii) ionic environment that directly affect DNA compaction (Iwataki et al, 2004). In the Δfis strain, DNA gyrases and Topo I actively change the DNA superhelicity without the barrier of Fis, and the topology of DNA turns into the dynamic state, in which the expression of Dps would easily results in the nucleoid condensation (Figure 8B). This is the case for S. aureus, whose MrgA expression and the nucleoid condensation are completely correlated (Morikawa et al, 2006). Database search showed that S. aureus lacks fis gene (Table I), but possess the orthologs of topA, gyrA and gyrB. Namely, the nucleoid dynamics in the wt strain of S. aureus correspond to those in the E. coli Δfis strain (Figure 8B). Taken together, the topology control of DNA seems to be critical for the Dps‐dependent nucleoid condensation.
A binding of Fis to DNA may physically block the interaction between DNA and Dps, because Fis binds to DNA in a sequence nonspecific manner (Schneider et al, 2001). However, since the amount of Fis in a cell in the log phase is ∼60 000 (Azam and Ishihama, 1999) and Fis functions as a homo‐dimer (Kostrewa et al, 1992), the amount of Fis may not be sufficient to physically block the entire genomic DNA from Dps (∼1 Fis dimer per 150 bp genome DNA, genome size=4.6 Mbp). In the Topo I++ and Gyrase++ cells with an ability of immediate nucleoid compaction upon H2O2 stimulation, the mRNA of fis was still present (Figure 6D). Therefore, the dynamics of nucleoid structure is likely controlled on the balance of the effects of Fis, Topo I, DNA gyrase and Dps, but not simply by a physical blocking (Figure 8C). The fact that the extracts from Topo I++ and Gyrase++ cell changed the topology of DNA supports this idea (Figure 7).
The relationship between DNA–Dps crystallization and nucleoid condensation
It has been reported that DNA–Dps interaction and the following DNA–Dps crystallization are independent of DNA superhelicity in vitro (Almiron et al, 1992; Wolf et al, 1999; Frenkiel‐Krispin et al, 2004). In vitro reconstitution of nucleosomes can also be done on a linear and relatively short DNA. However, although a nucleosome on a short linear DNA winds ∼146 bp of DNA, the efficiency on a long (∼100 kb) DNA is strongly depending on the superhelicity of the DNA (Hizume et al, 2004, 2005). In this study, we are dealing with the entire genome (∼4 Mbp). Therefore, the implication obtained from in vitro experiments may not be simply applicable to a much larger in vivo system, and thus, it is speculated that the topology control may be critical for the DNA–Dps crystallization when it comes to the entire genome.
The concentration of doubly charged cations has been known to contribute to the DNA–Dps interaction and crystallization (Grant et al, 1998; Frenkiel‐Krispin et al, 2001), being particularly important as a regulating signal for DNA–Dps interaction upon entry to the stationary phase (Hurwitz and Rosano, 1967). Therefore, such divalent cations are also expected to be critical for Dps‐dependent nucleoid condensation on the balance of the topological state of genome DNA. Since the overexpression of Dps in the log phase of wt never exposes the DNA–Dps crystalline state, the cation‐facilitated conformational change of the genome DNA might be required for bridging the DNA–Dps crystalline to the genome condensation towards stationary phase. Although at present, unfortunately, little information about the concentration of divalent cations is available for the log phase, toward the stationary phase, the genome DNA in E. coli is restructured and formed toroidal morphology inside the Dps crystal in a cell (Frenkiel‐Krispin et al, 2004).
Different response to oxidative stress via Dps induction in bacteria
We have previously reported that the dps gene is present widely in the bacterial kingdom (Kim et al, 2004; Takeyasu et al, 2004), and that it is generally induced in response to oxidative stresses through different transcription factors (Morikawa et al, 2006). One of them is OxyR, which is present in gamma‐Proteobacteria including E. coli, beta‐Proteobacteria, alpha‐Proteobacteria and Actinobacteria (Table I). Another is PerR, which is functional counterpart of OxyR and works as a repressor for the mrgA gene (Horsburgh et al, 2001, 2002; Morikawa et al, 2006). PerR is distributed in Firmicutes including S. aureus, Spirochete, Cyanobacteria, Hyperthermophilic bacteria, alpha‐Proteobacteria and Chlamidia (Table I). Since there is no primary‐sequence homology between PerR and OxyR, these two regulatory systems seem to have evolved independently. We have previously speculated that bacterial species that possess the dps gene have an ability to compact their nucleoid under oxidative stress, since S. aureus expresses MrgA under oxidative stress and condenses its nucleoid (Morikawa et al, 2006). However, in this study, we found that, due to the presence of Fis, the nucleoid in E. coli could not be condensed under oxidative stress that induced the Dps expression. In this case, timely expression of Dps as the Fe‐chelater (or ferroxidase) may have a role in inhibiting the production of hydroxyl radicals under oxidative stress.
We re‐compared the distributions between Fis and the transcription regulators of the dps gene in bacterial kingdom in terms of the nucleoid condensation sensitive to the oxidative stress. In order to precisely predict the actual involvement of OxyR and PerR, we applied the Schneider's information theory (Hengen et al, 1997; Schneider, 1997; Zheng et al, 2001). We explored the binding sites of OxyR and PerR on the promoter regions of dps and dps homologs throughout the bacterial kingdom (Table I). A comparison of the binding sites with their corresponding regulators revealed that the regulation system of dps by OxyR exists only in gamma‐Proteobacteria including E. coli, although OxyR itself is distributed in gamma‐Proteobacteria, alpha‐Proteobacteria and Actinobacteria. The distribution of Fis is also restricted in gamma‐Proteobacteria, suggesting that the Dps induction by oxidative stress in gamma‐Proteobacteria, in general, does not lead to a nucleoid condensation. In contrast, the PerR system is distributed not only in Bacillales including S. aureus but also in Lactobacillales, delta/epsilon‐Protebacteria, Fusobacteria, Spirochete and Cyanobacteria, indicating that the regulation of dps expression by PerR is more universal in bacteria and, therefore, seems to be the older system than by OxyR. Since the distribution of PerR system never overlaps with the presence of Fis, the species possessing the PerR system will induce an immediate nucleoid condensation under oxidative stress. From this sense, it is likely that the protection system of genome DNA against oxidative stress without nucleoid condensation have been developed specifically in gamma‐Proteobacteria, coupled with the later acquisitions of Fis and the OxyR dependent regulation of the dps gene.
Materials and methods
Bacterial strain and growth condition
wt (W3110) and the dps deletion mutant (W3110 (Δdps::Km)) are E. coli K‐12 derived strain (Kim et al, 2004). The oxyR, himD (ihf‐B) and fis deletion mutants (K‐12 BW25113 derived strains) were obtained from the KO collection (systematic knock out strain of E. coli K‐12; GenoBase: http://ecoli.aist‐nara.ac.jp/) by Baba et al (2006). P1 vir phages were prepared from these deletion mutants as donors. Thereafter deletion mutants of W3110 (ΔoxyR::Km, ΔhimD::Km and Δfis::Km) were constructed by P1 general transduction using the recipient strain W3110 (wt). Glycerol stocks of W3110 strains (wt, Δdps, ΔoxyR, ΔhimD (ihf‐B) and Δfis) were inoculated into LB medium and cultured at 37°C with constant shaking (180 r.p.m., Bioshaker BR‐15 (TAITEC)) for 24 h. Ten microliters of the saturated culture were inoculated into 5 ml of fresh LB medium and cultured at 37°C with constant shaking (180 r.p.m., Bioshaker BR‐15) to an appropriate cell density. The cell density was determined by measuring the absorbance at 600 nm by UV‐160A (Simadzu).
Western blot analysis
Each strain was grown in LB medium until the optical density (OD600) reached at 0.5 (log phase) and 2.0 (stationary phase), and was exposed to 2 mM H2O2 for 15 min. After 15‐min incubation at 37°C, the cells were harvested by a centrifugation at 13 000 g for 30 s at 4°C, suspended in an SDS sample buffer (50 mM Tris–HCl pH 6.8, 2% SDS, 1% 2‐mercaptoethanol, 10% glycerol, 0.025% bromophenol blue). An equal amount of the cell lysate measured by OD600 was subjected to SDS–PAGE and analyzed by immunoblotting with a polyclonal antibody against Dps. Specific binding was detected with ECL‐labeled secondary antibody.
Northern blot analysis
Every strain was grown in LB medium until the OD600 reached at 0.5 (log phase), and was exposed to the 2 mM H2O2. After 15 min incubation, the cells were harvested by a centrifugation at 13 000 g for 30 s at 4°C, suspended in a lysis buffer containing 10 mM Tris–Cl (pH 8.0), 10 mM EDTA and 400 μg/ml of lysozyme, and, then, incubated at 37°C for 2 min. The total RNA was extracted from the lysate by the SV total RNA isolation system (Promega). Four micrograms of the total RNAs was separated on a 1% agarose‐formamide denaturing gel and transferred onto the Hybond N+ membrane (Amersham Biosciences). The DNA fragments prepared by PCR from the W3110 genomic DNA were labeled by a random priming method using the Ready‐to‐Go DNA labeling Beads (Amersham Biosciences), and used as the probes. The hybridization was conducted at 60°C in the hybridization solution containing 5 × SSPE, 5 × Denhardt's solution, 0.5% SDS and 20 μg/ml of salmon sperm DNA for 16 h, and the final washing was carried out at 60°C in 0.1 × SSC and 0.1% SDS for 30 min.
‘On‐substrate lyses’ procedure
Bacterial cells were harvested from 100 μl culture by centrifugation (13 000 g, 1 min at 4°C) and washed once with 1 ml phosphate‐buffered saline (PBS) (pH 7.2). The cells were resuspended in 250 μl of PBS and a 50 μl aliquot was placed onto a round‐shape cover glass of 15 mm in diameter. The extra liquid was removed by nitrogen gas blow. Each strain was immersed in 2 ml of a buffer containing 10 mM Tris–HCl (pH 8.2), 1 mM NaN3 and 0.1 M NaCl for 5 min, followed by an addition of 25 μg/ml lysozyme. After 2 min incubation at 25°C, Brij 58 (polyoxyethylene hexadecyl ether) and sodiumdeoxycholate were added to the final concentrations of 0.25 and 0.1 mg/ml, respectively. After 10 min, the cover glass was dried under nitrogen gas. The surface of the cover glass was gently washed with distilled water and dried again for AFM analyses. For the observation by fluorescent microscopy (Axiobert 200, Olympus), the sample was stained by DAPI.
The atomic force microscope (SPI3800N‐SPA400) from Digital Instrument was used for the imaging of E. coli nucleoid structures in air at room temperature under a tapping mode with a 100 μm scanner. Probes made of a single silicon crystal with the cantilever length of 129 μm and the spring constant of 33–62 N/m (OMCL‐AC160TS‐W2, Olympus) were used for imaging. Data were collected in the height mode with a scanning rate of 0.5–1.0 Hz and the driving amplitude of 40–80 mV. The images were captured in a 512 × 512 pixels format and the captured images were flattened and plane‐fitted before analysis. The image analyses were performed with the software accompanying with the imaging module.
All of the AFM images contain ‘tip effect’. The sizes of the objects in the images were estimated at the half‐maximum height (FWHM, full‐width at half‐maximum) for correction of the tip effect (Schneider et al, 1998). In the present study, we have found that the AFM cantilevers purchased from Olympus had a constant tip angle of ∼35°, and a tip radius of 5±2 nm by measuring the apparent width of double‐stranded DNA in the AFM images.
Overexpression of Dps, Topo I and DNA gyrase in the wt and the Δfis strain
The coding sequences for the Dps, Topo I, DNA gyrase subunit A and subunit B was amplified by PCR and subcloned into the pCA24N vector (28), which contained the sequences for the green fluorescent protein (GFP) and 5xHis. The GFP sequence in the pCA24N vector was deleted by SalI restriction enzyme. The dps, topA, gyrA and gyrB genes are now under the control of the pT5/lac promoter. For the co‐expression of gyrA and gyrB in W3110, gyrB gene was subcloned into RSFDuet‐1 (Novagen). The gene expressions of topA and gyrA/B can be induced by IPTG. The plasmid DNA was then transformed into a fis‐deletion strain and the wt of W3110. The cells at different growth phases were exposed to IPTG (0.5 mM) for varying periods of time and subjected to Western blot and AFM analyses.
Topology assay of plasmid DNA
Cytosol extracts were prepared as following. Twenty‐five milliliters of bacterial cultures in the log phase (OD600, 0.5) were harvested by centrifugation at 5000 g for 10 min, and the bacterial pellets were resuspended in 1 ml buffer P (PBS+1 mM PMSF and protein inhibitor cocktail (Nakarai)). Lysozyme was added to the final concentration of 25 μg/ml, and incubated on ice for 30 min. After centrifugation at 5000 g for 10 min, glycerol (final concentration is 5%) was added to the supernatants (cell extracts), and the extracts were stored at −20°C. The amount of proteins in the cell extracts was determined by DC protein assay kit (Lowly method, BIO‐RAD).
pBluescript SKII‐(pBSII) was purified from the stationary phase culture (overnight culture) by midi‐prep kit (Qiagen). The cell extracts were added to 500 ng pBSII and incubated at 37°C for 15 min. The reaction was stopped by the addition of 1% SDS. Following phenol/chloroform extraction, the plasmid sample was collected by ethanol precipitation, resuspended in TE with 1 μg/ml RNase, and loaded into a 1.2% agarose gel containing 0.25 μg/ml chloroquine (Shure et al, 1977). Electrophoresis was performed in TAE buffer containing 0.25 μg/ml chloroquine for 15–17 h at 25 V in a refrigerator. The gels were washed with distilled water for 4 h, stained with EtBr for 2 h, and observed by FAS‐II (TOYOBO).
Information analysis of the IHF, OxyR and PerR binding sites on the dps gene promoter
Fifty‐one and nineteen known OxyR and PerR binding sites were collected, respectively, and each individual information weight matrix in a set of the binding sites was constructed according to the Schneider's information theory (Schneider, 1997). Then the matrix was applied to the promoter regions (−200∼ −1) of 76 dps genes in 68 bacterial species, which had been known to posses the dps gene (Morikawa et al, 2006), and the information contents in each promoter were calculated. The scores in Table I represent the highest scores in a certain dps promoter.
Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).
Supplementary Figure 1
Supplementary Figure 2
Supplementary Figure 3
This study was supported by the Special Co‐ordination Funds, the COE Research Grant and the Basic Research Grant (B) from the Ministry of Education, Culture, Sports, Science and Technology of Japan. We thank the Japan Science Society (the Sasakawa Scientific Research Grant) and the Sumitomo Foundation for their strong support for this work. We also thank Mr Junta Tsutsumi and Ms Shizuka Iwasaka for their technical supports in bioinformatics and molecular biology, respectively.
- Copyright © 2006 European Molecular Biology Organization